Discussion about this post

User's avatar
The Green Hornet's avatar

Peter Doshi at the BMJ called this out from the beginning. He demanded to see the actual data.

Expand full comment
Robert Yoho, MD's avatar

STATISTICAL TRICKERY: RELATIVE VS ABSOLUTE RISK IN MEDICAL STUDIES

Study authors and the media often report results in terms of RELATIVE rather than ABSOLUTE numbers, which vastly overstate the advantages of a therapy. Here is how it works.

To understand this, consider a medication that cuts heart attack rates in half, from 2 percent a year to one percent. This RELATIVE risk improvement sounds great, but it is misleading and some would say fraudulent. The all-important ABSOLUTE risk improvement—decrease in heart attacks for everyone who gets the therapy—is only one percent. Only one percent of those who took the drug benefitted.

The characteristics of the group that receives therapy is critically important. If the patients who are treated have severe heart disease with a heart attack rate of 25 percent instead of two percent a year, cutting this in half would be consequential. The ABSOLUTE improvement is still 50 percent but the RELATIVE improvement is 12.5 percent when the 25 percent heart attack rate is cut in half.

From Butchered by "Healthcare"

Expand full comment
30 more comments...

No posts

Ready for more?